gcastle_gae

GAE

Package

gCastle

Version

1.0.3

Language

Python

Docs

Paper

Ng et al.[1]

Graph type

DAG

Docker

bpimages/gcastle:1.0.3

Module folder

gcastle_gae

Description

A gradient-based algorithm using graph autoencoder to model non-linear causal relationships.

Example JSON

[
  {
    "id": "gcastle_gae",
    "x_dim": 1,
    "num_encoder_layers": 1,
    "num_decoder_layers": 1,
    "hidden_size": 4,
    "latent_dim": 1,
    "l1_graph_penalty": 0.0,
    "use_float64": false,
    "learning_rate": "1e-3",
    "max_iter": 10,
    "iter_step": 3000,
    "init_iter": 3,
    "h_tol": "1e-8",
    "init_rho": 1.0,
    "rho_thres": "1e+30",
    "h_thres": 0.25,
    "rho_multiply": 2.0,
    "early_stopping": false,
    "early_stopping_thres": 1.0,
    "graph_thres": 0.3,
    "timeout": null
  }
]